JP4370409B2 - Cancer prognosis prediction method - Google Patents

Description

  The present invention relates to a method for detecting the presence or absence of a pathogenic mutation in the p53 gene in a tumor tissue / tumor cell based on a gene expression profile obtained from the expression level of a gene in a cancer tissue sample of a cancer patient, and a method for predicting cancer prognosis And a kit that can be used in those methods.

  Cancer is a genetic disease and each tumor type has been shown to have several common molecular mechanisms for carcinogenesis. As a result of intensive research and development on the measurement of gene and protein expression profiles in recent years, it has become possible to measure the expression status of genes and proteins in cancer tissues. . Many methods for predicting the prognosis of cancer patients by applying these measurement methods have been reported, but so far, methods for predicting the prognosis easily and accurately for many types of cancer have been reported. It is also well known that this is not done.

  Genomic changes in tumors are diverse, and it is not easy to analyze individual tumors at the genetic level and elucidate the complete molecular mechanism of carcinogenesis. There are two approaches to elucidating the molecular mechanism of this complex cancer. One is a research method based on a simplified model using cultured cells and animal models. Although this method is extremely useful for elucidating the molecular mechanism and molecular pathway in the early stages of carcinogenesis, it is difficult to comprehend the entire complex system, and in the case of animal models, it is not always applicable to humans. is there.

Another method is to collect complex information exhaustively using human tumors and extract common molecular mechanisms from complex systems. Microarray data analysis methods can be broadly divided into two types: Unsupervised analysis and Supervised analysis. Unsupervised analysis classifies samples according to the similarity of gene expression patterns and seeks to find the correlation between each divided group and known histopathological factors and prognosis. Florin et al. Performed microarray on 36 breast cancers and found genes highly correlated with each of estrogen receptor (ER), lymph node metastasis, tumor diameter and clinical stage by Unsupervised analysis (Non-patent Document 1). . In addition, Sorlie et al. Showed by unsupervised analysis that breast cancers were divided into five groups according to gene expression patterns, and that each group had a different prognosis (Non-patent Document 2). On the other hand, Supervised analysis is a method of extracting genes having a statistically significant difference in expression between two known groups and creating a gene list that can distinguish the two groups. Microarray analysis using supervised analysis has been reported to include the presence or absence of ER expression and the presence or absence of human epidermal growth factor receptor (HER2) expression. Among them, van't Veer et al. 70 genes were extracted from 78 cases of developing breast cancer according to the presence or absence of distant metastasis within 5 years (Non-patent Document 3). In addition, the prognosis prediction using the 70 gene indicates that the accuracy is higher than that of the prediction based on the clinicopathological factors so far (Non-patent Document 4).

When DNA is damaged, the cells can either go to apoptosis or be repaired. When an error occurs in the DNA repair process, a mutation occurs in the gene, and the mutation causes cell canceration with a high probability. Therefore, DNA repair genes (oncogenes, tumor-related genes), tumor suppressor genes, growth factors (growth) factors) and apoptosis genes.

  The p53 gene is a tumor suppressor gene found in most human cancers and has an activity to stop the cell cycle. In particular, it is thought to work to stop the cycle of cells damaged by radiation, drugs, etc., or to induce apoptosis. Therefore, if there is an abnormality in the p53 gene, it is considered that damaged cells proliferate as they are and cancer occurs. P53 is a transcriptional regulator functioning as a tetramer, and activates a plurality of transcriptions such as p21 which is an inhibitory protein of CDK (cyclin dependent kinase) / cyclin complex involved in cell cycle control.

In Patent Document 1, the recurrence of cancer is determined by comparing the gene and / or protein expression pattern with the gene and / or protein expression pattern of a human primary tumor derived from a patient with cancer recurrence and a patient with no cancer recurrence. We propose a method to predict
Patent Document 2 shows that the prognosis of breast cancer can be predicted with high accuracy by measuring the expression states of 56 types of genes, but the prognosis for breast cancer is limited. Moreover, in patent document 3, although it proposes performing the diagnosis of cancer by measuring the state of p53 produced by p53 gene which is a tumor suppressor gene, the proposal regarding the prediction of prognosis is not performed. There have been many reports on p53 gene mutations so far, but it has been reported that 5-10% of reported mutations have errors in detection and notation errors (Non-Patent Documents 5 and 6). Therefore, in order to create a gene expression file that predicts the presence or absence of mutations in the p53 gene, it is necessary to maximize detection of p53 gene mutations and to minimize false negatives and false positives.
Special table 2004-536610 JP2004-344171 Special table 2001-523441 Selaru, FM, Yin, J., Olaru, A., Mori, Y., Xu, Y., Epstein, SH, Sato, F., Deacu, E., Wang, S., Sterian, A., Fulton, A., Abraham, JM, Shibata, D., Baquet, C., Stass, SA, and Meltzer, SJ An unsupervised approach to identify molecular phenotypic components influencing breast cancer features.Cancer Res, 64: 1584-1588, 2004. Sorlie, T., Perou, CM, Tibshirani, R., Aas, T., Geisler, S., Johnsen, H., Hastie, T., Eisen, MB, van de Rijn, M., Jeffrey, SS, Thorsen , T., Quist, H., Matese, JC, Brown, PO, Botstein, D., Eystein Lonning, P., and Borresen-Dale, AL Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications.Proc Natl Acad Sci USA, 98: 10869-10874, 2001. van 't Veer, LJ, Dai, H., van de Vijver, MJ, He, YD, Hart, AA, Mao, M., Peterse, HL, van der Kooy, K., Marton, MJ, Witteveen, AT, Schreiber, GJ, Kerkhoven, RM, Roberts, C., Linsley, PS, Bernards, R., and Friend, SH Gene expression profiling predicts clinical outcome of breast cancer.Nature, 415: 530-536, 2002. van de Vijver, MJ, He, YD, van't Veer, LJ, Dai, H., Hart, AA, Voskuil, DW, Schreiber, GJ, Peterse, JL, Roberts, C., Marton, MJ, Parrish, M ., Atsma, D., Witteveen, A., Glas, A., Delahaye, L., van der Velde, T., Bartelink, H., Rodenhuis, S., Rutgers, ET, Friend, SH, and Bernards, R. A gene-expression signature as a predictor of survival in breast cancer.N Engl J Med, 347: 1999-2009, 2002. Soussi, T., Kato, S., Levy, P., and Ishioka, C. Reassessment of the TP53 mutation database in human diesase by data mining with a library of TP53 missense mutations.Hum Mutat, 2004. Soussi, T., Dehouche, K., and Beroud, C. p53 website and analysis of p53 gene mutations in human cancer: forging a link between epidemiology and carcinogenesis.Hum Mutat, 15: 105-113, 2000.

  Methods for determining the presence or absence of mutations in the p53 gene include immunohistochemical staining, SSCP, and DNA sequencing. Immunohistochemical staining and SSCP have many false positives and false negatives, and are inferior to DNA sequencing in terms of mutation reliability. The superiority of DNA sequencing is evident in reports on correlation with prognosis. However, clinically, when using the presence or absence of mutations in the p53 gene as a prognostic factor, the DNA sequencing method requires sequencing of the entire translation region in order to accurately determine the mutation status. Takes time and effort. In addition, it is clear that the effect of mutation of the p53 gene cannot be determined only by the presence or absence of mutation. For this reason, in order to distinguish between the pathological mutations described below and non-pathological mutations, it is also necessary to evaluate the functions of the mutants found, and it is considered difficult to apply the DNA sequencing method clinically.

  The present invention overcomes the above problems, detects the presence or absence of pathological mutations in the p53 gene in tumor tissue / tumor cells, and receives a number of patients who receive unnecessary treatment for various types of cancer patients. It is an object of the present invention to provide a highly sensitive method for predicting a prognosis that makes it possible to determine an optimal prognosis treatment policy such as reducing the prognosis.

  As a result of diligent investigations in view of the above points, the present inventors pay particular attention to a specific gene that is affected by the expression level depending on the presence or absence of a pathogenic mutation in the p53 gene, which is a tumor suppressor gene. The wild type (low risk) group for a wide variety of cancer patients by finding the power gene group (the gene group consisting of 70 shown in Table 1) and measuring changes in the expression level of those genes and / or proteins Or, it was found possible to detect a pathological variant (high risk) group. In the present invention, even if the mutation is in the nucleotide sequence of the p53 gene, a missense that causes loss of transcriptional activity as shown in “Mutation analysis of p53 gene by DNA sequencing” in the present specification. Mutations other than "pathological mutations" such as mutations, protein-cutting mutations (nonsense mutations, frameshift mutations) and splicing site mutations are referred to as "non-pathological mutations". Included in “wild type”.

This invention is made | formed based on said knowledge, and is as follows.
That is, as a first aspect of the present invention, the following steps:
(1) An arbitrary gene group in Table 1 in a tumor tissue / tumor cell (meaning “a gene group consisting of 70 identified by the numbers in Table 1,“ Genbank ”and“ Common name ”). The same)),
(2) Phase between the gene expression profile obtained from the expression level and the gene expression profile of the pathogenic variant and the wild type (hereinafter also including “non-pathological variant”) of the p53 gene determined in advance Find the number of relationships
(3) A method for detecting the presence or absence of a pathogenic mutation in the p53 gene in the tumor tissue / tumor cell, comprising assigning the gene expression profile obtained to the gene expression profile of the type with the larger correlation coefficient .

As a second aspect of the present invention, the following steps:
(1) Measure the expression level of any gene group in Table 1 in tumor tissue / tumor cells,
(2) Obtaining a correlation coefficient between the gene expression profile obtained from the expression level and the previously determined pathogenic variant and wild type gene expression profiles of the p53 gene,
(3) Assign the obtained gene expression profile to the gene expression profile of the type with the larger correlation coefficient,
(4) A method for predicting the prognosis of cancer, comprising determining that the prognosis is poor in the case of a gene expression profile of a pathological variant and that the prognosis is good in the case of a wild type gene expression profile. Related.

Furthermore, the third aspect of the present invention relates to a kit that can be used in the method of the present invention, comprising a polynucleotide or oligonucleotide consisting of at least a part of the base sequence of any gene group in Table 1.

According to the present invention, the presence or absence of a pathological mutation of the p53 gene in a tumor tissue / tumor cell can be detected in a relatively short time and with high accuracy by using various means. The degree of prognosis of a patient can be predicted statistically significantly, and the prognostic treatment policy can be determined based on them.

  The gene group used in the method of the present invention is composed of genes arbitrarily selected from the genes listed in Table 1 (all 70 types). Although the number of genes to be selected is not particularly limited, the gene group is preferably selected from, for example, 9 or more, 10 or more, 25 or more, 33 or more, 46 or more, or all 70 genes. Composed.

  From Table 1, the selection method of each gene constituting the gene group used in the method of the present invention is also arbitrary. For example, it can be selected at random, or the P value is less than 0.05 based on the P value indicating the significance of the difference in expression level in the pathological variant expression profile and the wild type expression profile. Alternatively, it can be selected from genes smaller than 0.01. Furthermore, it is possible to select an appropriate number of genes at appropriate intervals in descending order of P value. As typical examples of the pathogenic variant expression profile and the wild type expression profile used in the method of the present invention, those shown in Table 7 can be mentioned.

  Furthermore, in the method for predicting the prognosis of cancer according to the present invention, in the step of determining whether the obtained expression profile belongs to a pathological variant expression profile or a wild type expression profile, depending on the magnitude of the correlation coefficient. When the P value in the correlation coefficient is 0.05 or more, the prediction accuracy can be further improved by making the cancer prognosis unpredictable.

Gene expression refers to the production of a gene product having the function of RNA or protein based on information held by the gene itself. Therefore, the expression level of a gene can be determined by measuring the amount of mRNA transcribed from the gene or the protein translated from the mRNA. In the method of the present invention, the expression level of the gene in the tumor tissue / tumor cell can be measured by any method / means known to those skilled in the art.

For example, measurement of the amount of mRNA expressed includes DNA microarray (or DNA chip), oligonucleotide microarray, Northern blotting (Northern hybridization), in situ hybridization, RNase protection assay, and reverse transcription polymerase chain reaction ( RT-PCR) and the like can be used. On the other hand, examples of the protein amount measurement include Western blotting, ELISA assay, protein array, immunohistochemical staining, and Yeast Two Hybrid.
Expression level analysis is performed by comparing such signal intensities. It is best to generate a ratio matrix of “expression intensity of gene in test sample” versus “expression intensity of gene in control sample”.
The present invention is based on finding an expression profile of a specific gene that can predict the prognosis of a cancer patient with high accuracy by statistically processing the expression level obtained by an existing measurement method. .
Gene expression profiles can also be displayed in various ways. The most common method is to arrange the ratio matrix into a graphical dendogram where the columns represent the test sample and the rows represent the genes. The data is arranged so that genes showing similar expression profiles are in close proximity to each other. The expression ratio of each gene is visualized by color. For example, a ratio less than 1 (indicating down-regulation) may appear in the blue portion of the spectrum, and a ratio greater than 1 (indicating up-regulation) may appear in the red portion of the spectrum. Such data can be displayed using commercially available computer software such as “GENESPRING” from Silicon Genetics, Inc. and “DISCOVERY” and “INFER” from Partek, Inc.

  Therefore, the kit of the present invention can take an appropriate configuration according to the method / means for measuring the expression level of the gene and / or the protein encoded by the gene in the tumor tissue / tumor cell. For example, the length of a polynucleotide or oligonucleotide consisting of at least a part of the base sequence of any gene in Table 1 included in the kit of the present invention can be several tens of base pairs, and specific parts thereof (bases) The sequence can be appropriately selected, designed and prepared by those skilled in the art based on information that can be easily obtained from various databases such as the database (Genbank) described in Table 1. Further, they can be used in the form of a DNA chip or a probe in Northern blotting, a primer in PCR or the like. Furthermore, if necessary, the polynucleotide or oligonucleotide may be labeled with an appropriate labeling substance such as a radioactive substance, a fluorescent substance, or a dye.

  The kit includes other elements or components known to those skilled in the art, such as various reagents, enzymes, buffers, reaction plates (containers), and the like, depending on the configuration and purpose of use.

  The target cancer or tumor tissue / tumor cell in the present invention is not particularly limited, and specifically includes head and neck cancer, stomach cancer, colon cancer, rectal cancer, liver cancer, gallbladder / bile duct cancer, pancreatic cancer, Examples include solid cancers such as lung cancer, breast cancer, bladder cancer, prostate cancer, and cervical cancer, and blood cancers such as malignant lymphoma and leukemia.

  EXAMPLES Hereinafter, although an Example demonstrates this invention in detail, the technical scope of this invention is limited and is not interpreted at all by description of a following example. Further, unless otherwise specified, the following examples are carried out by standard methods known to those skilled in the art, and the contents of the references cited in the present specification are the contents of the disclosure of the present specification. Part of it.

(Target)
Forty-two cases of invasive ductal cancer that were operated on breast and endocrine surgery at Tohoku University Hospital from March 2000 to November 2002 (with the approval of the Ethics Committee of the Graduate School of Medicine) (Formal consent was obtained). The background of all 40 patients is shown in Table 2. Age ranged from 30 to 78 years (average 58.45 years), 17 before menopause and 23 after menopause. ER was positive 30 cases, negative 10 cases, progesterone receptor (PR) was positive 35 cases, negative 5 cases. Of the positive cases, there were 25 cases with a total score of 5 or more (moderate positive) in both ER and PR. Of the negative cases, those with a total score of 0 were ER for 8 cases and PR for 4 cases. As for HER2, 0+ was 7 cases, 1+ was 16 cases, 2+ was 11 cases, and 3+ was 6 cases.

(Method)
Collection and storage of breast cancer tumor tissue The tumor tissue used was a part of the tumor tissue removed by surgery, and was rapidly frozen in liquid nitrogen and stored at -80C. The remaining tumor tissue was fixed with 10% formalin for histopathological examination and embedded in paraffin.

Creating frozen sections
The frozen tissue stored at −80 ° C. was embedded in White Tissue-Coat (UI Kasei Co., Ltd.) and frozen in n-hexane cooled with dry ice. The embedded tissue was sliced to a thickness of 20 μm with a freezing microtome and attached to a slide glass with a foil (Leica) coated with Poly-L-Lysine solution (Sigma). After fixing with ice-cooled ethanol / acetic acid = 19: 1 solution for 3 minutes, it was washed with ice-cooled diethylpyrocarbonate water (hereinafter, DEPC water) for 1 minute. After staining with ice-cooled 0.1% Toluidine blue (Chroma) for 3 minutes, it was washed twice with ice-cooled DEPC water again and dried with cool air from a dryer. The prepared frozen tissue sections were stored at −80 ° C. until microdissection.

Microdissection <br/> laser microdissection system frozen tissue section prepared: were microdissected at (Leica AS LMD system Leica, Inc.). Cut out the clumps of tumor cells with few stroma and inflammatory cells in the section with a laser, and for RNA extraction use 50 μl of buffer RLT (including 2-Melcaptoethanol 1%) of RNeasy Micro Kit (Qiagen) For DNA extraction, it was collected in 50 μl Buffer ATL of DNeasy Tissue Kit (Qiagen).

RNA extraction
Total RNA was extracted from tumor tissue (about 2 × 10 ^4-5 cells) microdissected using RNeasy Micro Kit (Qiagen). It was confirmed that the extracted RNA was not decomposed by the area ratio of the ribosomal RNA peaks of 18s and 28s using Bioanaraza (Agilent).

DNA extraction
Similar to RNA, genomic DNA of tumor cells was extracted from microdissected tumor tissues (5 × 10 ^3-4 cells) using DNeasy Tissue Kit (Qiagen). About 20 to 200 ng of genomic DNA was extracted from each specimen.

Mutation analysis of p53 gene by DNA sequencing Sequence analysis of p53 gene exon 4 to exon 8 was performed using genomic DNA of tumor cells extracted by microdissection. After evaluating the polymorphism of codon 72 in exon 4, if there were other nucleotide sequence changes, it was evaluated as follows according to the type of pathological mutation. That is, if a missense pathogenic mutation is found, its pathological significance is identified by a known transcription function evaluation database (Kato, S., Han, SY, Liu, W., Otsuka, K., Shibata, H., Kanamaru, R , and Ishioka, C. Understanding the function-structure and function-mutation relationships of p53 tumor suppressor protein by high-resolution missense mutation analysis.Proc Natl Acad Sci USA, 100: 8424-8429, 2003) When the activity was lost, it was regarded as “pathological mutation”. Since protein-cutting pathogenic mutations (nonsense pathological mutations, frameshift pathological mutations) and splicing site pathogenic mutations are translated, p53 is predicted to lack a DNA-binding domain or oligomerization domain. This mutation was also referred to as a “pathological mutation”.

In cases where no pathogenic mutation was found in exons 4-8, additional sequence analysis of exons 2, 3, 9-11 was added to cover the entire translation region. Using 2 to 10 ng genomic DNA of each sample, the composition of the reaction solution is 250 nM each primer, 250 μM dNTP, 1.5 U Ex Taq (TaKaRa), 25 mM TAPS Buffer, 50 mM KCl, 2 mM MgCl ₂ , 1 mM 2-Melcaptoethanol, total PCR reaction was performed with 30 μl. Exons 2, 3, 10, and 11 were subjected to PCR reaction, and then 1 μl of the PCR reaction solution was amplified again by the nested PCR method (each primer used is shown in Table 3, reaction time, reaction temperature, and cycle number are shown in Table 4). Pointing out toungue).

After the PCR reaction, purification was performed using GFX PCR DNA and Gel Band Purification Kit (Amersham Bioscience), and then a sequencing reaction was performed. The sequencing reaction was performed by adding 0.5 μM primer, 4.0 μl of DTCS Quick Start Master Mix (CEQ ^™ DTCS-Quick Start Kit, Beckman Coulter), and 100-300 ng of DNA after PCR amplification to create a 10 μl reaction system. Using System 9700 (Biosystems), heat denaturation at 96 ° C for 30 seconds, followed by heat denaturation: 96 ° C for 20 seconds, annealing: 50 ° C for 20 seconds, extension reaction: 60 ° C for 4 minutes for a total of 40 cycles went. The CEQ ^™ 2000XL DNA Analysis system (Beckman Coulter) was used as the automatic fluorescent DNA sequencer.

Immunohistochemical staining The tumor tissue fixed with paraffin was subjected to immunohistochemical staining of p53 using Monoclonal Mouse Anti-Human p53 Protein (Biomeda). A 1.5μm thick section was prepared from each specimen, deparaffinized with xylene, treated with endogenous peroxidase (10% H ₂ O _{2 in} methanol), washed with distilled water, p53 in a microwave oven, etc. Was pretreated in an autoclave. Both were heated using a citric acid buffer (10 mM citric acid, pH 6.0) in a microwave oven for 15 minutes and in an autoclave at 121 ° C. for 5 minutes. After washing with PBS and blocking nonspecific reaction, each primary antibody was reacted at 4 ° C. overnight. After the primary antibody reaction, it was washed with PBS (3 times for 5 minutes), and reacted with the secondary antibody for 30 minutes. Histfine kit (Nichirei) was used as the secondary antibody, but only PR (Chenicon) used ENVISION + (Dako) (in the histofine kit, the reaction was performed by adding 10% human serum to the secondary antibody. After washing with PBS (5 minutes 3 times), the reaction with streptavidin was performed for 30 minutes.) After washing again with PBS (5 minutes 3 times), color development (DAB 30 mg + 10 % Sodium azide 667 μl + 50 mM Tris Buffer (pH 7.6)). Thereafter, the plate was washed with PBS and stained with hematoxylin solution. Hercep Test was performed according to the attached manual.

The determination was performed as follows. p53 was positive if it stained in more than 10% of all nuclei in the section (Kandioler-Eckersberger, D., Ludwig, C., Rudas, M., Kappel, S., Janschek, E., Wenzel , C., Schlagbauer-Wadl, H., Mittlbock, M., Gnant, M., Steger, G., and Jakesz, R. TP53 mutation and p53 overexpression for prediction of response to neoadjuvant treatment in breast cancer patients. Res, 6: 50-56, 2000; and Stal, O., Stenmark Askmalm, M., Wingren, S., Rutqvist, LE, Skoog, L., Ferraud, L., Sullivan, S., Carstensen, J. , and Nordenskjold, B. p53 expression and the result of adjuvant therapy of breast cancer. Acta Oncol, 34: 767-770, 1995).

Northern hybridization
(1) Preparation of control RNA pool A control RNA pool was prepared by mixing equal amounts of total RNA extracted from each sample.
(2) Labeling of gene probes Each gene probe was labeled with ³² P-dCTP using Rediprime II DNA labeling system (Amersham Bioscience).
(3) Total RNA extracted from the RNA pool to be hybridized and each sample was heated in 2.2M formaldehyde, 50% formamide, 1x MOPS buffer solution at 65 ° C for 15 minutes, then 1% agarose gel (2.2M formaldehyde, 1x Electrophoresis was performed using MOPS buffer). The mixture was shaken in a 50 mM NaOH solution for 25 minutes and then shaken twice with 200 mM sodium acetate solution for 20 minutes, and then transferred to Hydrobond N + (Amersham Bioscience) for 12 hours. The solution was allowed to stand on a filter paper containing 50 mM NaOH solution for 5 minutes for alkali fixation and washed with 2 × SSC solution. Then, it was dried at 80 ° C. for 2 hours, and UV cross-linking was performed using a transilluminator.
The transferred membrane was immersed in a prehybridization solution (5x SSPE, 50% formamide, 5x Denhalt'solution, 0.5% SDS, 20 µg / ml salmon sperm DNA), and prehybridization was performed at 60 ° C for 1 hour. Thereafter, a labeled probe subjected to heat denaturation was inserted, and hybridization was performed at 60 ° C. for 17 hours.
After hybridization, wash with primary wash (2x SSPE, 0.1% SDS) for 10 minutes (room temperature) and secondary wash (1x SSPE, 0.1% SDS) for 20 minutes (60 ° C) twice each. It was. After washing with a tertiary washing solution (0.1x SSPE, 0.1% SDS) for 20 minutes (60 ° C), autoradiography was performed at -80 ° C. The results of autoradiography were digitized with a densitometer.

(result)
Detection of p53 gene mutation As a result of sequence analysis using exons 4 to 8 of the p53 gene, pathological mutations were found in 20 (50%) of all 40 cases (Table 5). Twenty cases of pathogenic mutations were found in 12 types of missense pathogenic mutations, a total of 13 cases (two of which were heterozygous), 3 nonsense pathological mutations, 2 frameshift pathological mutations (2 base deletions), There were 2 pathological mutations in the splicing site. Samples that were not found to have pathological mutations in exons 4 to 8 were further subjected to sequence analysis of exons 2, 3, and 9 to 11, but no new pathological mutations were found. All pathological mutations were mapped to DNA binding domains. The number of exons and cases in which pathogenic variants were found were 1 in exon 4, 6 in exon 5, 5 in exon 6, 7 in exon 7, and 1 in exon 8.

12 types of missense pathogenic mutations were confirmed by referring to the transcriptional function evaluation database of the comprehensive missense pathogenic mutation library (listed above), and all of them were loss-of-function mutations (pathological mutations) that lost transcription activation ability. . Proteolytic mutations (nonsense mutations, frameshift mutations) are apparently loss-of-function mutations due to the lack of a C-terminal tetramerization domain and a portion of the DNA binding domain. Two cases of splicing site mutations were predicted to be pathological mutations that skip exon 6 and exon 7, respectively. Since these exons encode part of the DNA binding domain, they were judged to be loss-of-function mutations (pathological mutations).

The results of mutational analysis in p53 gene by immunohistochemistry <br/> immunostaining of p53 was performed immunostaining p53 for the purpose of confirming the protein expression levels. As a result of determining that 10% or more of the whole tumor in the section was stained as positive, there were 14 positive cases and 25 negative cases (Table 6 and FIG. 1, results of 39 cases in total. 1 wild type case) Data is not available.) Of these, all missense mutations were positive, and all non-missense mutations were negative.

In general, the expression level of p53 is not controlled mainly by transcription or translation but by degradation, and even if cells are produced under normal conditions (under no stress), they are degraded more and more. Is hardly expressed. Only when the cells are under stress, the degradation of p53 is suppressed and expression as a protein is recognized. Missense mutant p53 accumulates in the cells by becoming less susceptible to degradation, and this result is recognized as positive on immunostaining. On the other hand, wild-type p53 under non-stress, pathological mutant p53 that has become non-responsive to anti-p53 antibodies due to protein-cutting mutations, etc. Therefore, it is expected to be negative.

Therefore, it was confirmed that the results of the immunostaining predicted by the missense mutation and the nonmissense mutation identified by the DNA sequence analysis described above completely coincided with each other.

Gene expression profile for detecting presence or absence of pathological mutation in p53 gene Table 70 shows 70 genes to be analyzed.
Using GAPDH (glyceraldehyde phosphate dehydrogenase) as an internal standard, the gene expression level between samples was standardized. The expression of each gene was calculated as the ratio of the expression level for each sample based on the expression level of the control RNA pool. Samples were divided into two groups: learning set for gene expression profile creation (26 cases) and test set for validation (12 cases). Two cases of heterozygous missense mutations were excluded from the analysis because there is a possibility that p53 function of cancer cells remains. Using the leaning set (12 pathological variant cases, 14 wild type cases), the average expression value of each type was calculated and used as the gene expression profile for each of the pathological variant and the wild type. Table 7 shows the expression ratio of the pathogenic mutant group as a pathogenic mutant expression profile in the leaning set, and the wild type group expression ratio as the wild type gene expression profile.

When the significance of the expression difference with and without pathological mutation was examined by Wilcoxon rank sum test, it was shown that all P values were less than 0.1. For these 70 genes, (a) all genes (70 genes), (b) genes with P values less than 0.05 (33 genes), (c) genes with P values less than 0.01 (10 genes), (d) P values (0.05) genes with a score of 0.05 or more, (e) 14 genes selected every 5 in the order of P values, (f) 10 genes from the largest P values, and (g) selected every 7 genes in the order of P values. (H) 10 genes selected every 7 in descending order of expression in the pathological variant group, and (i) 9 gene sets randomly selected 10 genes for each case. The correlation coefficient (Peason's correlation coefficient) with the gene expression profile of each of the mutant type and wild type was calculated, and based on the magnitude of the correlation coefficient, it was determined whether it was pathological mutant type or wild type.
The predictive predictive value of leaning set was 100% for all gene sets. Moreover, the predictive predictive value of the test set was 75.0% or more for all gene sets. In particular, the predictive predictive value of gene sets other than (g) to (h) was 88.3%.
From this result, it was confirmed that the pathogenic mutation in the p53 gene can be detected (predicted) with an accuracy of 75.0% or more by measuring the expression level of an arbitrary gene group selected from the 70 genes in Table 1 ( Table 8).

Examination of correlation between presence / absence of detected pathogenic mutation of p53 gene and prognosis (1) For the case of Example 1, the accuracy of prognosis prediction of the present invention was examined. That is, among the breast cancer patients in Table 2, 29 early stage breast cancer patients in stage I to IIB (out of 32, one who died of other diseases and 2 heterozygous mutations were excluded) were listed in Table 8 (b). Based on the expression pattern of 33 genes, we divided 15 patients into wild-type prediction group (low risk group) and 14 persons with pathological variant prediction group (high risk group), followed up and observed their recurrence free survival A Kaplan-Meier curve for the period was created (Figure 2). So far, all relapse cases are in the high-risk group and there are no relapse cases from the low-risk group. Moreover, it is roughly divided into a low risk group and a high risk group, and is ideally stratified. Furthermore, despite the short period of less than five years, the P value is very small at 0.0322, indicating the high accuracy of prognosis prediction.

(2) In general, prediction methods such as prognosis based on gene expression patterns are often not reproducible even by a third party due to variations in cell composition in the sample, variations in the sample pretreatment stage, and the like.
We decided to verify the reproducibility of this prediction method using third-party microarray data. That is, Van't Veer et al. Published microarray data for 78 patients with early breast cancer (Non-Patent Document 3). The published expression data of about 24,000 genes include 46 genes (genes marked with “◯” in Table 9 (a)) out of the 70 genes in Table 2.

Therefore, using their data, we determined whether each case belongs to the gene expression profile of the high-risk group (pathological variant) or low-risk group (wild-type), and for each type, the survival time of recurrence without distant metastasis Kaplan-Meier curves for (metastasis free survival) were generated (FIGS. 3A and 3B).
As shown in Table 9, the gene sets are (a) 46, (b) 25, (c) 9, (d) 21, (e) 7, (f) 10, (g 10) (h) 10 8 gene sets were used. As a result, in (a), (c), (f), (g), and (h), a statistically significant difference in prognosis was observed depending on the presence or absence of the predicted pathogenic mutation of the p53 gene. In addition, although (b), (d), and (e) were not statistically significant, the P value was as very small as 0.0532 to 0.0754. Despite the use of third-party data, it can be said that it shows excellent reproducibility.

  Furthermore, the correlation coefficient used for prediction with P ≧ 0.05 is considered not statistically significant, and the result when not predicted is shown in FIG. 3C. As a result, in the Kaplan-Meier curve, the not predict group was located between the pathological variant prediction group and the wild type prediction group, and the prognosis tended to be clearly divided. In (a) and (b), P = 0.0036 and 0.0024, respectively, which were statistically very significant results. In (c) and (d), the same tendency as in (a) and (b) was observed, but no statistically significant value was obtained as the P value. This is because the number of genes used for prediction is small, and the P value of the correlation coefficient used for prediction becomes large, the number of cases that become not predict increases, the pathological variant type prediction group, the wild type prediction group This is thought to be because the number of cases has decreased. For the same reason, prediction was impossible in (e).

3A-C, the horizontal axis indicates the number of months, and the vertical axis indicates the ratio of patients who have not experienced recurrence of distant metastasis (distant metastasis recurrence rate). In addition, the P values shown in these figures are the results of testing the difference in recurrence rate of no distant metastasis between the two groups by the log rank test, and statistically significant differences when the P value is less than 0.05. Is considered to be. The Kaplan-Meier curve is generally called a survival curve (usually because “death” is often the event) and is created using the Kaplan-Meier method (http: //www.medical-tribune) .co.jp / BENRI / survival2.htm etc.) In addition, “mt-predict” and “wt-predict” in each figure of FIG. 3 indicate a pathological mutation type prediction group and a wild type prediction group, respectively. The number of cases used in each prediction method is as follows.

FIG. 3A: (a) mt 42 wt 36, (b) mt 48 wt 30, (c) mt 44 wt 34, (d) mt 38 wt 40
FIG. 3B: (e) mt 43 wt 35, (f) mt 36 wt 42, (g) mt 39 wt 39, (h) mt 42 wt 36
3C: (a) mt 18 wt 29 np 31, (b) mt 15 wt 29 np 34, (c) mt 3 wt 6 np 69, (d) mt 12 wt 8 np 58

From these results, the presence or absence of pathological mutations in the p53 gene was detected, and the prognosis was clearly higher in the group designated as the high risk group (pathological variant) than in the group predicted as the low risk group (wild type). Was shown to be bad. Thus, it was demonstrated that the prognosis of cancer can be predicted using a gene group arbitrarily selected from Table 1.

In Example 1, detection of pathological mutations in the p53 gene by gene expression profile was performed by Northern hybridization. However, as already described, any other method, for example, microarrays as in van't Veer et al. It is possible to use. The microarray can detect pathological mutations in the p53 gene in a single experiment, and the required time is considered to be within several days until the result is obtained from the collection of the specimen. Considering that immunohistochemical staining also requires at least 2 days, it is considered that clinical application is sufficiently possible. In addition to Northern hybridization and microarray, any method that can perform semi-quantification of RNA, such as quantitative RT-PCR, can be used, and is considered highly versatile.

Furthermore, the detection of pathological mutations in the p53 gene based on the gene expression profile of the present invention allows not only the presence / absence of mutations in the p53 gene but also the evaluation of the pathological significance at the same time, compared with other methods. However, it is considered very useful. According to the present invention, in addition to the detection of the presence or absence of pathological mutations in the p53 gene, it is shown that the prognosis can be predicted by the gene expression profile, which is considered to be clinically useful data.

A number of mutations in the p53 gene and susceptibility to treatment have also been reported. Jasson et al. Examined mutations in the p53 gene and reported that radiation therapy improved prognosis in the mutant type, but no improvement in the wild type (Jansson, T., Inganas, M. , Sjogren, S., Norberg, T., Lindgren, A., Holmberg, L., and Bergh, J. p53 Status predicts survival in breast cancer patients treated with or without postoperative radiotherapy: a novel hypothesis based on clinical findings. Clin Oncol, 13: 2745-2751, 1995). Multiple variants have been reported to be less sensitive to tamoxifen and paclitaxel, which are commonly used in the treatment of breast cancer, and CEF therapy, which is a commonly used regimen. (Kandioler-Eckersberger, D., Ludwig, C., Rudas, M., Kappel, S., Janschek, E., Wenzel, C., Schlagbauer-Wadl, H., Mittlbock, M., Gnant, M., Steger, G., and Jakesz, R. TP53 mutation and p53 overexpression for prediction of response to neoadjuvant treatment in breast cancer patients.Clin Cancer Res, 6: 50-56, 2000; Berns, EM, Foekens, JA, Vossen, R ., Look, MP, Devilee, P., Henzen-Logmans, SC, van Staveren, IL, van Putten, WL, Inganas, M., Meijer-van Gelder, ME, Cornelisse, C., Claassen, CJ, Portengen, H., Bakker, B., and Klijn, JG Complete sequencing of TP53 predicts poor response to systemic therapy of advanced breast cancer.Cancer Res, 60: 2155-2162, 2000; Geisler, S., Lonning, PE, Aas, T., Johnsen, H., Fluge, O., Haugen, DF, Lillehaug, JR, Akslen, LA, and Borresen-Dale, AL Influence of TP53 gene alterations and c-erbB-2 expression on the response to treatment with doxorubicin in locally advanced breast cancer. Cancer Res, 61: 2505-2512, 2001.). However, treatment with an anticancer drug or radiation has not only a cancer cell but also a normal cell, and has various side effects. At present, the treatment method is determined by clinical stage and histopathological diagnosis, but due to the low accuracy of prognosis prediction, patients who have no recurrence after surgery have to be treated excessively. There is no situation. In addition, treatment that cannot be expected even in patients who need treatment should be avoided. By using the present invention, it becomes possible to determine a prognostic treatment policy accurately and quickly, which is very useful clinically. It is thought that.

1 is an immunohistological staining diagram (photo) in Example 1. FIG. It is a figure which shows the Kaplan-Meier curve in Example 2 (1). It is a figure which shows the Kaplan-Meier curve in Example 2 (2). It is a figure which shows the Kaplan-Meier curve in Example 2 (2). It is a figure which shows the Kaplan-Meier curve in Example 2 (2).

Claims (12)

The following steps:
(1) All 70 genes below in breast cancer tissue / breast cancer cells:
CENPF (SEQ ID NO: 29), ASPM (SEQ ID NO: 30), PKMYT1 (SEQ ID NO: 31), BIRC5 (SEQ ID NO: 32), HSPC150 (SEQ ID NO: 33), CDCA8 (SEQ ID NO: 34), ANAPC7 (SEQ ID NO: 35), UBE2C (SEQ ID NO: 36), PTTG1 (SEQ ID NO: 37), LOC51161 (SEQ ID NO: 38), TGS (SEQ ID NO: 39), BCL11A (SEQ ID NO: 40), PLK (SEQ ID NO: 41), CDC45L (SEQ ID NO: 42), CENPE ( SEQ ID NO: 43), PTP4A2 (SEQ ID NO: 44), C10orf3 (SEQ ID NO: 45), STMN1 (SEQ ID NO: 46), FLJ11280 (SEQ ID NO: 47), FLJ14399 (SEQ ID NO: 48), CCNB2 (SEQ ID NO: 49), FLJ10719 (sequence) No. 50), PRC1 (SEQ ID NO: 51), KIF2C (SEQ ID NO: 52), HIS1 (SEQ ID NO: 53), BF930764 (SEQ ID NO: 54), MUTYH (SEQ ID NO: 55), MGC45866 (SEQ ID NO: 56), MGC7036 (SEQ ID NO: 57), SULF2 (SEQ ID NO: 58), MAPRE1 (SEQ ID NO: 58) 59), MKNK2 (SEQ ID NO: 60), RPS27L (SEQ ID NO: 61), TMSNB (SEQ ID NO: 62), TTC12 (SEQ ID NO: 63), HCAP-G (SEQ ID NO: 64), CEAL1 (SEQ ID NO: 65), FLJ33962 ( SEQ ID NO: 66), GMNN (SEQ ID NO: 67), ENST00000332343 (SEQ ID NO: 68), HEC (SEQ ID NO: 69), GMPR2 (SEQ ID NO: 70), TncRNA (SEQ ID NO: 71), SMOC2 (SEQ ID NO: 72), DNAJC9 (sequence) No. 73), RAD54B (SEQ ID NO: 74), CKS2 (SEQ ID NO: 75), I_960269 (SEQ ID NO: 76), BAG1 (SEQ ID NO: 77), AL137566 (SEQ ID NO: 78), BRRN1 (SEQ ID NO: 79), CDC2 (SEQ ID NO: 80), ZF (SEQ ID NO: 81), THC1577090 (SEQ ID NO: 82), CDKN2C (SEQ ID NO: 83), I_1842252 (SEQ ID NO: 84), SDOS (SEQ ID NO: 85), SNAPC2 (SEQ ID NO: 86), EVI2A (SEQ ID NO: 87) ), V4b (SEQ ID NO: 88), BC007934 (SEQ ID NO: 89) ECT2 (SEQ ID NO: 90), RAD21 (SEQ ID NO: 91), MCM7 (SEQ ID NO: 92), AK097469 (SEQ ID NO: 93), MGC39900 (SEQ ID NO: 94), FLJ21439 (SEQ ID NO: 95), STATIP1 (SEQ ID NO: 96), DKFZp434L142 (SEQ ID NO: 97) and PLAT (SEQ ID NO: 98), or
Sequence number 29-34,36-38,40-47,49,51-53,55,59-62,64,67,69,70,72,74,75,77,78,80,81,83, 86, 87, 90 to 92, and the expression level of any 10 or more gene groups of 46 gene groups represented by 96, 98 ,
(2) Obtaining a correlation coefficient (Peason's correlation coefficient) between the gene expression profile obtained from the expression level and the previously obtained pathogenic variant and wild type gene expression profiles of the p53 gene,
(3) The gene expression profile obtained is assigned to the gene expression profile of the type with the larger correlation coefficient, provided that the p53 gene pathological mutation is P value in the correlation coefficient of 0.05 or more. A method for detecting the presence or absence of a pathological mutation of the p53 gene in the breast cancer tissue / breast cancer cell. Sequence number 29-34,36-38,40-47,49,51-53,55,59-62,64,67,69,70,72,74,75,77,78,80,81,83, The method according to claim 1, wherein any 25 or more genes in 46 gene groups represented by 86, 87, 90 to 92, 96 to 98 are used. Sequence number 29-34,36-38,40-47,49,51-53,55,59-62,64,67,69,70,72,74,75,77,78,80,81,83, The method according to claim 1, wherein any 33 or more genes in 46 gene groups represented by 86, 87, 90 to 92, 96 to 98 are used. The method according to claim 2 or 3, wherein the P value indicating the significance of the difference in expression level in the pathological variant expression profile and the wild type expression profile of all the genes used is less than 0.05. . Any one of the following gene groups (a), (c), (f), (g) and (h) selected from the 70 genes according to claim 1 :
(a) SEQ ID NOs: 29 to 34, 36 to 38, 40 to 47, 49, 51 to 53, 55, 59 to 62, 64, 67, 69, 70, 72, 74, 75, 77, 78, 80, 81 , 83, 86, 87, 90-92, 96-98, 46 gene groups;
(c) a group of nine genes represented by SEQ ID NOs: 29-34 and 36-38;
(f) 10 gene groups represented by SEQ ID NOs: 80, 81, 83, 86, 90 to 92, and 96 to 98;
(g) 10 gene groups represented by SEQ ID NOs: 29, 34, 41, 46, 53, 62, 72, 80, 90 and 98; and
(h) measuring the expression level of 10 gene groups represented by SEQ ID NOs: 32, 41, 51, 53, 55, 62, 64, 77, 80 and 98;
(2) Obtaining a correlation coefficient (Peason's correlation coefficient) between the gene expression profile obtained from the expression level and the previously obtained pathogenic variant and wild type gene expression profiles of the p53 gene,
(3) A method for detecting the presence or absence of a pathogenic mutation in the p53 gene in the breast cancer tissue / breast cancer cell, comprising assigning the gene expression profile obtained to the gene expression profile of the type with the larger correlation coefficient. The following steps:
(1) All 70 genes below in breast cancer tissue / breast cancer cells:
CENPF (SEQ ID NO: 29), ASPM (SEQ ID NO: 30), PKMYT1 (SEQ ID NO: 31), BIRC5 (SEQ ID NO: 32), HSPC150 (SEQ ID NO: 33), CDCA8 (SEQ ID NO: 34), ANAPC7 (SEQ ID NO: 35), UBE2C (SEQ ID NO: 36), PTTG1 (SEQ ID NO: 37), LOC51161 (SEQ ID NO: 38), TGS (SEQ ID NO: 39), BCL11A (SEQ ID NO: 40), PLK (SEQ ID NO: 41), CDC45L (SEQ ID NO: 42), CENPE ( SEQ ID NO: 43), PTP4A2 (SEQ ID NO: 44), C10orf3 (SEQ ID NO: 45), STMN1 (SEQ ID NO: 46), FLJ11280 (SEQ ID NO: 47), FLJ14399 (SEQ ID NO: 48), CCNB2 (SEQ ID NO: 49), FLJ10719 (sequence) No. 50), PRC1 (SEQ ID NO: 51), KIF2C (SEQ ID NO: 52), HIS1 (SEQ ID NO: 53), BF930764 (SEQ ID NO: 54), MUTYH (SEQ ID NO: 55), MGC45866 (SEQ ID NO: 56), MGC7036 (SEQ ID NO: 57), SULF2 (SEQ ID NO: 58), MAPRE1 (SEQ ID NO: 58) 59), MKNK2 (SEQ ID NO: 60), RPS27L (SEQ ID NO: 61), TMSNB (SEQ ID NO: 62), TTC12 (SEQ ID NO: 63), HCAP-G (SEQ ID NO: 64), CEAL1 (SEQ ID NO: 65), FLJ33962 ( SEQ ID NO: 66), GMNN (SEQ ID NO: 67), ENST00000332343 (SEQ ID NO: 68), HEC (SEQ ID NO: 69), GMPR2 (SEQ ID NO: 70), TncRNA (SEQ ID NO: 71), SMOC2 (SEQ ID NO: 72), DNAJC9 (sequence) No. 73), RAD54B (SEQ ID NO: 74), CKS2 (SEQ ID NO: 75), I_960269 (SEQ ID NO: 76), BAG1 (SEQ ID NO: 77), AL137566 (SEQ ID NO: 78), BRRN1 (SEQ ID NO: 79), CDC2 (SEQ ID NO: 80), ZF (SEQ ID NO: 81), THC1577090 (SEQ ID NO: 82), CDKN2C (SEQ ID NO: 83), I_1842252 (SEQ ID NO: 84), SDOS (SEQ ID NO: 85), SNAPC2 (SEQ ID NO: 86), EVI2A (SEQ ID NO: 87) ), V4b (SEQ ID NO: 88), BC007934 (SEQ ID NO: 89) ECT2 (SEQ ID NO: 90), RAD21 (SEQ ID NO: 91), MCM7 (SEQ ID NO: 92), AK097469 (SEQ ID NO: 93), MGC39900 (SEQ ID NO: 94), FLJ21439 (SEQ ID NO: 95), STATIP1 (SEQ ID NO: 96), DKFZp434L142 (SEQ ID NO: 97) and PLAT (SEQ ID NO: 98), or
Sequence number 29-34,36-38,40-47,49,51-53,55,59-62,64,67,69,70,72,74,75,77,78,80,81,83, 86, 87, 90 to 92, and the expression level of any 10 or more gene groups of 46 gene groups represented by 96, 98 ,
(2) Obtaining a correlation coefficient (Peason's correlation coefficient) between the gene expression profile obtained from the expression level and the previously obtained pathogenic variant and wild type gene expression profiles of the p53 gene,
(3) The gene expression profile obtained is assigned to the gene expression profile of the type with the larger correlation coefficient. However, if the P value in the correlation coefficient is 0.05 or more, the prediction of cancer prognosis is not possible. Exclude as possible ,
(4) A method for predicting the prognosis of breast cancer, comprising determining that the prognosis is poor in the case of a pathogenic variant gene expression profile and that the prognosis is good in the case of a wild type gene expression profile. Sequence number 29-34,36-38,40-47,49,51-53,55,59-62,64,67,69,70,72,74,75,77,78,80,81,83, The method according to claim 6, wherein any 25 or more genes in 46 gene groups represented by 86, 87, 90 to 92, 96 to 98 are used. Sequence number 29-34,36-38,40-47,49,51-53,55,59-62,64,67,69,70,72,74,75,77,78,80,81,83, The method according to claim 6, wherein any 33 or more genes in 46 gene groups represented by 86, 87, 90 to 92, 96 to 98 are used. The method according to claim 7 or 8, wherein the P value indicating the significance of the difference in expression level in the pathological variant expression profile and the wild type expression profile of all the genes used is less than 0.05. . Any one of the following gene groups (a), (c), (f), (g) and (h) selected from the 70 genes according to claim 6 :
(a) SEQ ID NOs: 29 to 34, 36 to 38, 40 to 47, 49, 51 to 53, 55, 59 to 62, 64, 67, 69, 70, 72, 74, 75, 77, 78, 80, 81 , 83, 86, 87, 90-92, 96-98, 46 gene groups;
(c) a group of nine genes represented by SEQ ID NOs: 29-34 and 36-38;
(f) 10 gene groups represented by SEQ ID NOs: 80, 81, 83, 86, 90 to 92, and 96 to 98;
(g) 10 gene groups represented by SEQ ID NOs: 29, 34, 41, 46, 53, 62, 72, 80, 90 and 98; and
(h) measuring the expression level of 10 gene groups represented by SEQ ID NOs: 32, 41, 51, 53, 55, 62, 64, 77, 80 and 98;
(2) Obtaining a correlation coefficient (Peason's correlation coefficient) between the gene expression profile obtained from the expression level and the previously obtained pathogenic variant and wild type gene expression profiles of the p53 gene,
(3) Assign the obtained gene expression profile to the gene expression profile of the type with the larger correlation coefficient,
(4) A method for predicting the prognosis of breast cancer, comprising determining that the prognosis is poor in the case of a pathogenic variant gene expression profile and that the prognosis is good in the case of a wild type gene expression profile. The polynucleotide or oligonucleotide consisting of a base sequence of several tens of base pairs of at least a part of each gene selected from the group of 70 genes according to claim 1. Kit that can be used in the method of The kit according to claim 11, wherein the polynucleotide or oligonucleotide is labeled.

Images